Atrial fibrillation (AF) affects over 2.2 million Americans, with projections of 12 million people by 2050. Stroke is the most serious complication of AF -15-20% of all strokes occurring in patients with AF - and results from thrombus formation in the left atrium (LA) and subsequent embolization. Stroke prevention in AF focuses on inhibiting thrombus formation with the use of warfarin in selected high risk patients. Although warfarin is effective at reducing stroke risk, it is also associated with a 3.4% annual risk of major bleeding. Clinicians utilize risk scores to identify patients who benefit from warfarn. However, an expert panel of the American College of Chest Physicians recently reported that the predictive value of these scores is mediocre, at best. This means that many patients who are not receiving warfarin will suffer a stroke while many patients who are treated with warfarin are being unnecessarily exposed to the risk of major bleeding. One reason for the subpar performance of these scores is that they are comprised of clinical risk factors (e.g. age > 75 years, high blood pressure) that are associated with stroke on a population basis but are not specifically related to thrombus formation in an individual based on physiologic principles. One of the key factors responsible for thrombus formation is blood stasis in the LA. Studies using ultrasound have linked markers of low flow in the LA to risk of stroke. However, this technique is not suitable for widespread screening and is unable to map 3D flow. The novel technique of functional 4D cardiovascular magnetic resonance imaging (4DcMRI) incorporating time- resolved 3D anatomy and 3D blood velocity information can assess 3D flow characteristics and is uniquely suited to evaluate blood stasis within the LA and right atrium (RA). We will apply thi technique to evaluate whether differences in stasis or flow patterns in the LA and RA appendages explain the approximately 10-fold greater risk of LA versus RA appendage thrombus formation and the 10-fold greater risk of systemic versus pulmonary embolism in AF. We hypothesize that stasis and regions with altered flow patterns are more common in the LA versus RA appendage in patients with AF at risk for intraatrial thrombus formation. We will a) optimize the 4DcMRI settings for flow characterization in AF patients; b) develop semi-automated analysis tools to calculate indices of stasis and assess 3D flow patterns; and c) compare flow patterns in two groups: 1) patients with persistent AF and documented prior atrial thrombus or thromboembolic stroke; 2) patients with persistent AF for >4 years not treated with warfarin and without history of LA thrombus or thromboembolic event. The hypothesis is that indices of stasis and flow pattern alterations will be significantly different in the LA versus RA appendage in Group 1 versus 2. This project will help determine whether 4DcMRI can be used to provide physiologic information regarding the risk for thrombus formation. The long-term goal is to evaluate whether individualized evaluation of indices of stasis can provide additional valuable information to advise which AF patients require warfarin therapy, thereby reducing both strokes and bleeding complications.